DRAFTING DEVICE FOR LEAK SURVEY SCHEDULING, DRAFTING
SYSTEM FOR LEAK SURVEY SCHEDULING, AND DRAFTING METHOD FOR
LEAK SURVEY SCHEDULING
5 BACKGROUND OF THE INVENTION
The present invention relates to a drafting device
for leak survey scheduling, and more particularly to a
drafting device for leak survey scheduling higher in cost
effectiveness.
10 Claim 1 of Japanese Patent Application Publication
No. 2011-59799 refers to: "A leak-surveyable route
extracting system is a system for analyzing collected and
accumulated past data on leak repairs in a water piping
network process for supplying clean water to end users,
15 comprising a master DB in which past data on leak repairs
and data on the number of underground water service pipes;
data extracting means for extracting data including the
number of repairs under each drawing number and the number
of underground water service pipes for each project number,
20 repair years and installation years; mesh evaluating means
for calculating, on the basis of the data obtained from
the data extracting means, the number of repairs and
repair ratios under each drawing number relative to the
whole water piping network, and numerically expressing or
2 5 em hat ' ca Lye p~ayfl-&hgy;at~-&at-ed-va-1-ues-gar-each- --
-- ~EQLD-EQ_.H-?:&:~-~-%-L- - - _ - -
drawing; project number list extracting means for
extracting project numbers belonging to drawing numbers of
higher leak risks obtained by the mesh evaluating means,
and listing project numbers, the number of repairs in
5 drawings, the total number of repairs, the number of
underground water service pipes and repair ratios; and
leak survey priority evaluating means for calculating, on
the basis of the extracted project number list, leak
survey priorities from user-set parameters."
10
SUMMARY OF THE INVENTION
The leak-surveyable route extracting system
according to Japanese Patent Application Publication No.
2011-59799, though it can extract from repair information
15 geographical areas of higher leak risks, cannot determine
the geographical area to be surveyed for any leak by
allocating limited (human and other) resources. Moreover,
the leak-surveyable route extracting system according to
Japanese Patent Application Publication No. 2011-59799
20 cannot address the uncertainty of leak risks.
In view of the problems noted above, the present
invention is intended to provide a drafting device for
leak survey scheduling higher in cost effectiveness even
where the limitation of available resources entails
To solve the problems, the configurations stated in
the Claims are used, for instance. The present patent
application includes a number of means for addressing the
problems, one of which is a drafting device for leak
s survey scheduling, drafting a leak survey scheduling to
cover a plurality of areas into which a water pipeline
network is partitioned, including a measurement
information collecting unit for collecting measurement
information pertaining to water flow rates from a
l o flowmeter installed on a water pipeline network and other
measuring instruments; a water consumption quantity memory
unit for storing information on water consumption
quantities in the areas; a leak quantity estimating unit
for estimating water leak quantities in the areas on the
15 basis of the measurement information and the water
consumption quantity information; a pipeline information
memory unit for accumulating pipeline information
including length information on the water pipeline network
in the areas; a survey and repair information memory unit
20 for accumulating survey and repair information including
the implementation timing of leak surveying and pipeline.
repairing in the areas; a prediction model learning unit
for generating prediction model information to predict the
trend of water leak quantities in the areas on the basis
pipeline information and the survey and repair
information; a leak quantity predicting unit for
generating predicted leak quantity information in the
areas on the basis of the prediction model information;
5 and a survey schedule drafting unit for drafting a leak
survey scheduling that prescribes the sequence of
implementing leak surveys in the plurality of areas on the
basis of the predicted leak quantity information, wherein
the leak quantity predicting unit generates predictions of
l o both the expected value of the leak quantity and the
uncertainty of the expected value of the leak quantity as
the predicted leak quantity information; and the survey
schedule drafting unit drafts scheduling by using a leak
cost calculated on the basis of both the expected value of
15 the leak quantity and the uncertainty of the expected
value of the leak quantity.
The invention has the advantageous effect of
providing a drafting device for leak survey scheduling
higher in cost effectiveness.
BRIEF DESCRIPTION OF THE DRAWINGS
Other problems, configurations and advantageous
effects than those described above will be made clear by
the following description of an embodiment of the
25 invention when taken in- o ' ct' -wi-th-thel-a~eompany-i.ng--
1 P O_B_ELH f -W7 - 7~l~~~~~l-$I~~~~:dL~
drawings, wherein:
Fig. 1 is a block diagram of a leak survey schedule
drafting device in this embodiment;
Fig. 2 is a hardware block diagram of the leak
5 survey schedule drafting device in this embodiment;
Fig. 3 shows a water piping network and water
distribution blocks for which the leak survey schedule
drafting device is to draft scheduling;
Fig. 4 shows tabulated pipeline information on water
lo distribution pipes recorded in a pipeline information
memory unit;
Fig. 5 shows tabulated pipeline information on water
service pipes recorded in the pipeline information memory
unit;
15 Fig. 6 shows tabulated leak survey information
recorded in a survey and repair history memory unit;
Fig. 7 shows tabulated pipeline repair information
recorded in the survey and repair history memory unit;
Fig. 8 shows an example of trend of an actual leak
20 quantity from a water distribution block;
Fig. 9 shows an example of prediction trend of a
leak quantity by a leak quantity predicting unit;
Fig. 10 shows tabulated prediction model information
to be recorded in a prediction model memory unit;
2 5, Fig 1 s ows u at su-r-vye- s ch edu-1-i-ng-- - - -
-1 P 0- _DIE-LH~ - ~ - $ ~ l % ~ l k f ~ ~ ~ ? f % k -
information to be recorded in a leak survey scheduling
memory unit;
Fig. 12 is a flow chart of processing by a
prediction model learning unit;
5 Fig. 13 is a flow chart of processing by a leak
quantity predicting unit;
Fig. 14 is a flow chart of processing by a survey
scheduling drafting unit;
Fig. 15 shows frame display by a frame display unit
lo regarding leak survey scheduling; and
Fig. 16 shows frame display by the frame display
unit regarding leak quantity prediction results.
I DETAILED DESCRIPTION OF THE INVENTION
15 An exemplary embodiment of the invention will be
described below with reference to drawings, in which the
same reference numerals are assigned to substantially the
same elements 'with no repetition of description.
Fig. 1 is a block diagram of a leak survey.schedule
20 drafting device 101 in this embodiment.
The leak survey schedule drafting device 101
includes a leak quantity estimating unit 110, a prediction
model learning unit 111, a leak quantity predicting unit
112, a survey schedule drafting unit 113, a pipeline
25 jnfqrmati m y=~~~-~2~-~ys&-~vey-and-~epa=i=rrhhi=~E~~y- - .2 7- -1 p if2 D - E . L H X -- - - - -
memory unit 122, a measurement information memory unit 123,
a water consumption quantity memory unit 124, a cost
information memory unit 125, a prediction model memory
unit 131, a survey scheduling memory unit 132, a survey
5 terminal IF unit 171, a measuring device IF unit 172, a
meter reading terminal IF unit 173, and a frame display
unit 174.
The leak survey schedule drafting system 100 has the
leak survey schedule drafting device 101, a survey
l o terminal 181, a measuring device 182 and a meter reading
terminal 183.
The configuration of the water piping network, water
distribution blocks and other elements for which the leak
survey schedule drafting device 101 is to draft scheduling
1 5 will be described afterwards with reference to Fig. 3.
The leak quantity estimating unit 110, to which the
measurement information recorded in the measurement
information memory unit 123 and the water consumption
quantity recorded in the water consumption quantity memory
20 unit 124 are inputted, estimates the leak quality in each
area in a prescribed period (one day for instance, and
transmits the estimated leak quality to the prediction
model learning unit 111 as leak quantity information.
Estimation of the leak quantity can be processed by
rate at night or water balance calculation (estimation of
the leak quantity by subtracting from the accumulated
quantity of distributed water the quantity of water
consumed and other factors). Regarding the leak quantity
5 information, which constitutes the output of the leak
quantity estimating unit 110, supplementary description
will be given with reference to Fig. 8.
The prediction model learning unit 111, to which
leak quantity in the area such as water distribution block
lo received from the leak quantity estimating unit 110,
pipeline information recorded in the pipeline information
memory unit 121 and survey and repair information recorded
in the survey and repair history memory unit 122 are
inputted, learns a prediction model for calculating
15 predictions of both the expected value of the future leak
quantity in each area and its uncertainty, and transmits
the learned prediction model to the prediction model
memory unit 131. Details of processing by the prediction
model learning unit 111 will be described afterwards with
20 reference to Fig. 12.
The leak quantity predicting unit 112, to which
prediction model information recorded in the prediction
model memory unit 131 is inputted, calculates predictions
of both the expected value of the future leak quantity in
the survey schedule drafting unit 113 and the frame
1 display unit 174. Details of processing by the 'leak
quantity predicting unit 112 will be described afterwards
with reference to Fig. 13.
5 The survey schedule drafting unit 113, to which leak
quantity prediction information received from the leak
quantity predicting unit 112 and cost information recorded
in the cost information memory unit 125 are inputted, ~
processes drafting calculation for leak survey scheduling
l o by using the leak cost calculated on the basis of
predictions of both the expected value of the leak
quantity and its uncertainty as evaluation indicators of
leak survey scheduling, and transmits the drafted leak
survey scheduling to the survey scheduling memory unit 132.
Leak survey scheduling here is what prescribes the
sequence of implementing leak surveys in a plurality of
areas. A leak survey means surveying by workers to find
out by acoustic examination, for instance, of water
service piping whether or not there is any problematic
20 leak. Details of processing by the survey schedule
drafting unit 113 will be described afterwards with
reference to Fig. 13.
The pipeline information memory unit 121 keeps in
record pipeline information including information on the
pipeline network for which the leak survey schedule
drafting device 101 is to draft scheduling. More
specifically, it keeps on record pipeline information on
water distribution piping and pipeline information on
5 water service piping, to be described afterwards with
reference to Fig. 4 and Fig. 5, respectively.
The survey and repair history memory unit 122 keeps
on record survey and repair information, including the
implementation timing of leak surveys and pipeline repairs,
lo regarding the water pipeline network for which scheduling
by the leak survey schedule drafting device 101 is to be
drafted. More specifically, there are on record leak
survey information and pipeline repair information to be
described afterwards with reference to Fig. 6 and Fig. 7,
15 respectively.
The measurement information memory unit 123 keeps on
record sensor-measured information in the water piping
network for which scheduling by the leak survey schedule
drafting device 101 is to be drafted. Sensor-measured
20 data (measured pressure and flow rate) of each measuring
device in a period transmitted by the measuring device 182,
such as one-minute period, are on record in a time series.
The water consumption quantity memory unit 124 keeps
on record information on water quantities consumed by city
25 water consumeArs_in the-w tAe~- g=networ-k-Eor-whi-eh-- - -
-- 1 Ea: -DI.LH-~_Q 7 - I~~EZIO:PZB~A-~%E
scheduling by the leak survey schedule drafting device 101
is to be drafted. Each water consumer receives water
distribution through a water service pipe from a water
distribution pipe of the water piping network, and a water
5 meter for charging the water consumption fee at the
receiving point. The water consumption quantity measured
by the water meter is periodically read by meter reading
personnel, an automatic meter reading system or the like.
The water consumption quantity memory unit 124 keeps on
10 record information on the quantities of water consumed by
individual city water consumers collected in this way,
namely information on the quantity of water consumed by
any given consumer in a given period.
The cost information memory unit 125 keeps on record
1s information on cost items including the costs of leak
surveys and pipeline repairs and losses from leaks in unit
quantity in the water piping network for which scheduling
by the leak survey schedule drafting device 101 is to be
drafted.
2 0 The costs of leak surveys and pipeline repairs in
this context mean the costs incurred when a leak survey by
acoustic detection or otherwise is made in each area of
the water piping network and pipeline repairs are made
against leaks found in the survey. The losses from leaks
consequences of leaving one cubic meter of water leak as
it is.
The prediction model memory unit 131 keeps on record
prediction model information outputted by the prediction
5 model learning unit 111. Specifics will be described
afterwards with reference to Fig. 10.
The survey scheduling memory unit 132 keeps on
record survey schedule information outputted by the survey
schedule drafting unit 113. Specifics will be described
lo afterwards with reference to Fig. 11.
The survey terminal IF unit 171 transmits to the
survey and repair history memory unit 122 information
received from the survey terminal 181 to be described
afterwards, and augments or updates leak survey
15 information and pipeline repair information on record in
the survey and repair history memory unit 122.
The measuring device IF unit 172 transmits to the
measurement information memory unit 123 information
received from the measuring device 182 to be described
20 afterwards, and adds new measurement information to the
measurement information memory unit 123.
The meter reading IF unit 173 transmits to the water
consumption quantity memory unit 124 information received
from the meter reading terminal 183 to be described
consumed water quantity on record in the water consumption
quantity memory unit 124.
The frame display unit 174 presents, to the operator
of the leak survey schedule drafting device 101, leak
quantity prediction information received from the leak
quantity predicting unit 112 and leak survey scheduling
information on record in the survey scheduling memory unit
132 through an output device, such as a display unit.
More specifically, the information is presented in
tabulated or graphic forms to be described with reference
to Fig. 15 and Fig. 16.
Each measuring device 182 is connected to the leak
survey schedule drafting device 101 via a communication
network. Measuring instruments installed in the water
pipeline network to be managed, including flowmeters and
pressure gauges, transmit the measured sensor data to the
measuring device IF unit 172 of the leak survey schedule
drafting device 101 via the communication network. A
specific example of measuring instrument will be described
afterwards with reference to Fig. 3.
Each survey terminal 181 and meter reading terminal
183 (mobile information terminal, such as POA) are
connected the leak survey schedule drafting device 101 via
the communication network.
pertinent survey terminal 181 to which information on leak
survey and pipeline repair is inputted as manipulated by
the user of the survey terminal, and the information is
transmitted to the survey terminal IF unit 171 of the leak
5 survey schedule drafting device 101.
Further the meter reading terminal 183, at the time
of meter reading to charge for water consumption, the
reading of the water meter inputted by the user of the
water meter to transmit the information to the meter
10 reading IF unit 173 the leak survey schedule drafting
device 101.
Fig. 2 is a hardware block diagram of the leak
survey schedule drafting device 101 in this embodiment.
Referring to Fig. 2, the leak survey schedule drafting
15 device 101 has a CPU 201, a memory 202, a media
input/output unit 203, a communication control unit 204,
an input unit 205, a display unit 206, a peripheral
equipment IF unit 207 and a bus 210.
The CPU 201 executes programs on the memory 202.
20 The memory 202 temporarily stores programs, tables and the
like. The media input/output unit 203 is an interface
with information recording media such as SD cards.
The communication control unit 204, connected to a
network 220, performs communication interfacing with
25 external- ~p&r us T e, ut u.ni-t-20~-is-a-user- -- --------
'IFa -LDGZL-H I-p &-~~fk:y!l&-~~d-~
interface comprising a keyboard, mouse and so forth.
The display unit 206 is a display described with
reference to Fig. 1. The peripheral equipment IF unit 207
is an interface with apparatuses positioned nearby,
5 including a printer.
The bus 210 mutually connects the CPU 201, the
memory 202, the media input/output unit 203, the
communication control unit 204, the input unit 205, the
display unit 206 and the peripheral equipment IF unit 207.
10 As comparison of Fig. 1 and Fig. 2 reveals, the leak
survey schedule drafting device 101 of Fig. 1 is operated
by execution of a program by the CPU 201.
Fig. 3 shows a water piping network and water
distribution blocks for which the leak survey schedule
15 drafting device 101 is to draft leak survey scheduling.
In Fig. 3, a distribution reservoir 301 from which water
is supplied to the water pipeline network and a
distribution pipeline network, represented by solid lines,
are illustrated. Also, flowmeters 310 to 313 and pressure
20 gauges 321 to 324 are shown as measuring instrument
installed in the water pipeline network.
The water distribution block means any region, among
the regions into which the water pipeline network is
divided, all the distribution pipes for water flowing into
flowmeters. In the case shown in Fig. 3, a flowmeter 311
and flowmeters 312 and 313 are installed on the respective
flow-in pipe paths of regions 331 and 332 to constitute
water distribution blocks. Incidentally, a water
5 distribution block may sometimes be referred to as a
district metered area (DMA) .
The survey schedule drafting unit 113 of the leak
survey schedule drafting device 101 drafts, as leak survey
scheduling, a schedule prescribing the sequence of leak
l o surveying of the water distribution block or areas into
which the water distribution block is further divided. In
the following description, the unit of leak surveying
including the water distribution block will be referred to
as an area.
15 The leak survey or surveying in this context means
human work to detect any water leak from water piping by
using hardware items such as an acoustic bar, some other
acoustic device and/or a correlational leak detector. A
specific case of leak survey scheduling will be described
20 afterwards with reference to Fig. 11.
Fig. 4 shows tabulated pipeline information 400 on
water distribution pipes recorded in the pipeline
information memory unit 121. The tabulated pipeline
information 400 has columns of distribution pipe ID
or ati p os bti o;fa-llI& fo dlga-Li0 n*4 02 -&l-neg kh- ---- ---
1.p B;.-dk!f -#x 8?-"Pk'1-hh : 2- -
information 403, diameter information 404, pipe type
information 405, life-in-service information 406,
accessory item information 407 and area information 408.
In the tabulated pipeline information 400 column, the leak
5 survey schedule drafting device 101 keeps on record all
available information on all the distribution pipes of the
water pipeline network under management, each row
representing one distribution pipe. Fig. 4 shows only one
distribution pipe by way of example.
10 The distribution pipe ID information 401 stores IDS
each of which uniquely identifies a specific distribution
pipe among all the distribution pipes. The positional
information 402 stores coordinate information that
identifies the installed position of a given distribution
15 pipe in linkage with the Geographic Information System
(GIs). The extension information 403 and the diameter
information 404 respectively store the length and diameter
of a given distribution pipe.
Information items including the pipe type (ductile
20 iron pipe (DIP) or cast iron pipe (CIP) ) and
specifications of the pipe (such as the presence or
absence of anti-corrosive sleeve) are stored in the pipe
type information 405. The number of years of the
distribution pipe in service is recorded into the life-inmay
be recorded therein, from which the length of service
can be calculated. Information items including the
numbers and positions of hydrants, air valves and other
accessory items and the number of branching-out water
5 service pipes are stored into the accessory item
information 407. Information on the area (water
distribution block) to which the pertinent distribution
pipe belongs is stored into the area information 408.
Fig. 5 shows tabulated pipeline information 500 on
lo water service pipes recorded in the pipeline information
memory unit 121. The tabulated pipeline information 500
keeps in its columns water service pipe ID information 501,
positional information 502, connectable distribution pipe
ID information 503, length information 504, diameter
15 information 505, pipe type information 506, life-inservice
information 507 and area information 508. The
tabulated pipeline information 500 keeps on record all
available information on the water service pipes of the
water pipeline network under management by the leak survey
20 schedule drafting device 101, with each row of the table
representing one service pipe. Fig. 5 shows only one
service pipe by way of example.
The water service pipe ID information 501 stores IDS
each of which uniquely identifies a specific water service
information 502 stores coordinate information that
identifies the installed position of a given water service
pipe. The connectable distribution pipe ID information
503 stores the ID of the distribution pipe to which the
5 pertinent water service pipe is connected.
The length information 504 and the diameter
information 505 respectively store the length and diameter
of a pertinent water service pipe. Information items
including the pipe type (polyethylene (PE) pipe and lead
l o pipe (LP) are stored in the pipe type information 506.
The number of years of the water service pipe in service
is recorded into the life-in-service information 507. Or
the year of its installation may be recorded therein.
Information on the area (water distribution block) to
15 which the pertinent water service pipe belongs is stored
into the area information 508.
Fig. 6 shows tabulated leak survey information 600
indicating leak survey information recorded into the
survey and repair history memory unit 122. The tabulated
20 leak survey information 600 has in its columns survey ID
information 601, period information 602 and object area
information 603. In the tabulated leak survey information
600, all the information on the history of leak surveys
implemented in all the areas under management by the leak
25 surve sc dul _ dr g dyye-4.g 1 - is-recor-ded7eaeh-r-ow- -- - - --
IP_O_ D E L $ I - & ? > ~ I I ~ ~ ~ ~ & ~ - - - -
representing leak surveys of one area. In Fig. 6, only
two leak survey histories are shown by way of example.
The survey ID information 601 stores IDS each of
which uniquely identifies a specific survey history out of
5 all the specific survey histories. Into the period
information 602, information on the period in which the
pertinent survey was implemented is stored. Into the
object area information 603, the ID of the area in which a
leak survey was implemented during the pertinent survey is
l o stored.
Fig. 7 shows tabulated pipeline repair information
recorded in the survey and repair history memory unit 122.
Tabulated pipeline repair information 700 has in its
columns repair ID information 701, type information 702,
15 pipe ID information 703, positional information 704, cause
information 705, prevented leak quantity information 706,
day and hours information 707 and survey ID information
708. The tabulated pipeline repair information 700 keeps
on record all information on the pipeline repair history
20 of the distribution pipes and water service pipes under
management by the leak survey schedule drafting device 101,
each row representing one pipeline repair. In Fig. 7, one
pipeline repair history is shown by way of example.
The repair ID information 701 stores IDS each of
C - n' ly'de 'fi- e , i &c-~epa=&r-h=i=s kor-y-out-of-
- X.FE~-D%-Lf ,~%&Y - xL25b-A~z3:~~-6 - - --
all the repair histories. Into the type information 722
and the pipe ID information 703, information indicating
whether a given object of repair is a distribution pipe or
a water service pipe and ID information on the object pipe.
5 Into the positional information 704, information on the
position of implementation of the pertinent is stored in
in linkage with the GIs. Namely, information on what
position of the object pipe has been repaired.
The cause information 705 stores information on the
lo presumable causes of repaired leaks. The causes are
identified as, for instance, deterioration overtime,
corrosion, overload or unknown. The prevented leak
quantity information 706 stores information on the result
of on-site estimation the leak quantity (flow rate) of the
15 repair object at the time of leak repair. The day and
hours information 707 stores information on the day and
hours of implementation of the pertinent repair. The
survey ID information 708 stores ID information on the
leak survey during which the leak to be repaired was
20 discovered. If the leak repaired is not the leak
discovered during the leak survey, but a leak discovered
by a notification of leak at the ground level for instance,
ID information that can identify the reason of the
discovery is stored.
i g . s w s-a x mpJ-e--f0 brend-o f-an-ae-kua=l-l-eak---------------------- -
- P P ~ - D FId~ f 4 8_rP2tq_sB:%d-abab3Ae
quantity from a water distribution block. In the graph of
Fig. 8, the horizontal axis represents time and the
vertical axis, the leak quantity from a given area (water
distribution block). Generally, the leak quantity in an
5 area monotonously increases as indicated by the leak
quantity 801 except in the period of execution of leak
preventing work represented by a preventive work period
811 and the preventive work period 812. The leak
preventing work in this context means leak survey and
lo repair of any leak discovered by the leak survey.
The leak quantity estimating unit 110, as stated
above, estimates the past leak quantity in each area, on
the basis of information on record in each memory unit, as
shown in Fig. 8 for instance.
15 Fig. 9 shows an example of prediction trend of a
leak quantity by the leak quantity predicting unit 112.
Referring to Fig. 9, predictions of both the expected
value of the leak quantity in a given area (water
distribution block) by the leak quantity predicting unit
20 112 and its uncertainty will be described. In the graph
of Fig. 9, the horizontal axis represents the length of
time since the last leak survey in the pertinent area and
the vertical axis, the leak quantity from the pertinent
area (water distribution block) .
he predi t r~ 8-o n-~-Qf - ~2 ioa~ ehnd=-~~ddh~L-~1~-ea k=quant-i=&y-ou&pu&t--ed - --
by the leak quantity predicting unit 112 pertains to three
time series corresponding to the length of time having
elapsed since the last leak survey and repair, including
that of an expected predictable leak quantity value 901, a
5 low predictable leak quantity value 903 and a high
predictable leak quantity value 902. The high predictable
leak quantity value 902 and the low predictable leak
quantity value 903 are values for expressing the
uncertainty of the expected predictable leak quantity
lo value 901. The high value and the low value can be, for
instance, the upper limit and the lower limit of a 95%
reliable section figured out from the degree of fitness of
the prediction model to the past trend of leak quantity.
Usually, since many uncertain factors are involved
15 in the occurrence of leak, it is difficult to treat any
leak as a decisive factor. For this reason, prediction
merely of the expected predictable leak quantity cannot
I
take into account differences in the degree of prediction I
uncertainty from area to area.
2 0 The leak survey schedule drafting device 101 can
I
perform more useful prediction of leak quantities by
I
outputting the high value and the low value as uncertainty
indicators of the expected quantity in drafting more
effective leak survey scheduling. As the method of I
I
h - ncer ad- -y= . -k~Gexpected-quant=iit=yY7t-he ~- -- -_-I I P H ~yg:mL%-fLel~Ls-$i_$:_$:~~%I -
magnitude of deviation from the expected quantity, for
instance, may as well be used instead of the high value
and the low value.
Fig. 10 shows tabulated prediction model information
5 1000 to be recorded in the prediction model memory unit
131. The tabulated prediction model information 1000 has
columns of prediction model ID information 1001, object
area information 1002, model type information 1003,
explanation variable information 1004, coefficient
l o information 1005 and type information 1006.
The tabulated prediction model information 1000
keeps on record all the prediction models outputted by the
prediction model learning unit 111, each row representing
one prediction model. In Fig. 10, prediction model are
15 shown. To add, the tabulated prediction model information
1000 includes at least one prediction model per area
applicable to all the areas that the leak survey schedule
drafting device 101 may take up as the object of schedule
drafting.
2 0 The prediction model ID information 1001 stores IDS,
each uniquely identifying a prediction model. The object
area information 1002 stores IDS of areas to which a
prediction model can be applied. Regarding a model
applicable to a plurality of areas, information that can
25 ident ' -I? li- c&a -&f$ ~jy%&K~tOredd_-- - --_
- xe_o_-aEb - _ .- -
The model type information 1003 stores the reference
numbers of numerical expressions for use by prediction
models in.making prediction. The explanation variable
information 1004 stores information on explanation
5 variables of areas to be used in applying prediction
models to different areas. The prediction model learning
unit 111 and the leak quantity predicting unit 112
acquires, as required, specific values of the explanation
variables of the pertinent area from the pipeline
l o information memory unit 121 and the survey and repair
history memory unit 122. The explanation variables will
be further described afterwards.
The coefficient information 1005 stores specific
values of coefficients of a prediction formula indicated
15 by the model type information 1003. The type information
1006 stores information indicating whether the pertinent
prediction model is a generic prediction model or an areaspecific
prediction model.
The prediction model learning unit 111 outputs
20 prediction model information described with reference to
Fig. 10. The prediction models outputted by the
prediction model learning unit 111, as described with
reference to the type information 1006, are classified
into two types including area-specific prediction models
L le - - - 1- -+I.& o~ek~"-&~a~di9-+6s~9=~i+=Q f~fic%-a-=re3a-Lan d-a-gener-i-c-
- -.
prediction models that can be made applicable to a
plurality of areas by substituting values in the area
described above.
For instance, the prediction model whose prediction
5 model ID is E4A332 shown in Fig. 10 is an area-specific
prediction model. As a specific example of prediction
formula of an area-specific prediction model, the leak
survey schedule drafting device 101 uses the following
prediction formula for instance. In the following
l o description, formulas (la) to (lc) will be generically
ref erred to as formulas (1) .
L(t) =LO + k x t . . . (la)
~h(t) = L(t) + DhO + mh x t . . . (lb)
Ll(t) = L(t) - DIO - ml x t . . . (lc)
15 where
L(t) is the expected value of predicted leak quantity [in
m3/hl ,
LO is the initial value (immediately after leak repair) of
the same [in m3/h],
20 Lh(t) is the high value of predicted leak quantity [in
m3/hl ,
DhO is the initial value (immediately after leak repair)
of the same [in m3/hl,
LI(t) is the low value of predicted leak quantity [in
DIO is the initial value (immediately after leak repair)
of the same [in m3/hl,
k, mh and ml are positive coefficients, and
t is the number of days having elapsed.
5 For instance, the prediction model whose prediction
model ID is E3AGE shown in Fig. 10 is a generic prediction
model. As a specific example of prediction formula of a
generic prediction model, the leak survey schedule
drafting device 101 uses the following prediction formula
l o for instance. In the following description, formulas (2a)
to (2c) will be generically referred to as formulas (2) .
~ ( tx,) = LO(X) + k(x) x t . . . (2a)
~h(t,x ) = L (t, x) + D ~ O+ mh x t . . . (2b)
LI(t, x) = L (t, x) - DIO - mIO x t ... (2c)
15 [Formula 11
Lo (x) = exp (a, + Csa, x x,) . . . (2d)
[Formula 21
where I
20 L(t, x) is the expected value of predicted leak quantity
[in m3/hl ,
LO(X) is the initial value (immediately after leak repair)
of the same [in m3/h1,
Lh(t, x) is the high value of predicted leak quantity [in
DhO is the initial value (immediately after leak repair)
of the same [in m3/hl,
LI(t, x) is the low value of predicted leak quantity [in
m3/hl
5 DIO is the initial value (immediately after leak repair)
of the same [in m3/hl ,
a , Po , as and ps are coefficient,
k(x), mh and ml are positive coefficients,
s is the index of explanation variables,
l o x s is the index s of explanation variables,
x is the explanation variable of all explanation variables,
and
t is the number of days having elapsed.
For s, the indexes of all the explanation variables are
15 taken.
As explanation variables of generic prediction
models handled by the leak survey schedule drafting device
101, any desired indicators that can be figured out from
the information stored in the pipeline information memory
20 unit 121, the survey and repair history memory unit 122
can be used, such as the number of leak repair cases in
the pertinent area, that of leak repair cases discovered
not by leak surveying but by notification (the number of
notified leak repair cases), the number of water service
the pipes longest in service, for instance 30-year or
older pipes.
If a different set of explanation variables are used
even if the same prediction formula ( 2 ) as the above is
5 used, the leak survey schedule drafting device 101 will
use them as another different prediction model. To add,
generic prediction models to be recorded in the prediction
model memory unit 131 may include not only the models
outputted by the prediction model learning unit 111 but
l o also generic prediction models voluntarily added by the
user of the leak survey schedule drafting device 101.
For instance, a generic prediction model learned in
an area not covered by schedule drafting by the leak
survey schedule drafting device 101 may be stored into the
15 prediction model memory unit 131 in advance, and the leak
quantity predicting unit 112 may use that generic
prediction model for prediction of the leak quantity in
the area covered by schedule drafting by the leak survey
schedule drafting device 101. The voluntarily added.
20 generic prediction model may as well fix the model
regarding the expected value, and the prediction model
learning unit 111 may be caused to learn coefficients
concerning the uncertainty, namely in the case of
prediction formula (2) only DhO, DIO, mh and ml.
information 1100 to be recorded in the leak survey
scheduling memory unit 132. The leak survey scheduling,
as described already, is to determine the sequence of
implementing a leak survey in a plurality of areas. For
5 instance, in addition to determining the sequence of
implementation, the period of implementing a leak survey
in each individual area can be determined. However, some
areas may be made exceptions to leak surveying. The leak
survey scheduling information 1100 is supposed to
l o determine the period of implementing a leak survey in each
area, and indicates for each area indicated by area ID
information 1101 which out of scheduled periods 1102 to
1104 a leak survey is to be implemented.
For instance Fig. 11 shows that a leak survey will
15 be implemented three times in a three-year schedule in an
area 331, namely in a survey period 1111, a survey period
1112 and a survey period 1113, and twice in an area 332,
namely in a survey period 1121 and a survey period 1122.
The leak survey scheduling information 1100 stores this
20 survey schedule for every area covered by the schedule.
Fig. 12 is a flow chart of processing by the
prediction model learning unit 111. In Fig. 12, there is
charted an operation flow in which the prediction model
learning unit 111 makes extractions on the basis of
at' %p 2r- 5ai -r~,p -&~g~----~~a~h.f~o~ded -a-rea ~f-U-r-&her I
conducts learning to generate a prediction model and
transmits the generated model to the prediction model
memory unit 131. At a start step 1200, the prediction
model learning unit 111 begins processing.
5 At a reception step 1201 for inputted information,
the prediction model learning unit 111 receives pipeline
information on record in the pipeline information memory
unit 121 and survey and repair information on record in
the survey and repair history memory unit.122 and leak
l o quantity information in each area from the leak quantity
estimating unit 110.
At a water distribution district extraction step
1202, the prediction model learning unit 111 extracts one
water distribution district (area), which is the source of
15 learning for the area-specific prediction model. However,
the prediction model learning unit 111 extracts only those
area in which the trend of past leak quantity is estimated
by the leak quantity estimating unit 110.
At an area-specific prediction model learning step
20 1203, the prediction model learning unit 111 learns areaspecific
prediction models for the extracted areas.
In order to learn an area-specific prediction model
having the prediction formula (1) as its prediction
formula, the prediction model learning unit 111 calculates
actual data of the leak quantity 801 during the preventive
work period 811 and the preventive work period 812 of Fig.
8 received from the leak quantity estimating unit 110.
For this calculation, a known technique such as least
5 squares can be utilized.
Also, the prediction model learning unit 111 can
determine, as a technique to determine the initial values
DhO and DIO and the coefficients mh and ml by using the
prediction formula (la) figured out as described above for
l o instance, the smallest initial values DhO and DIO and the
coefficients mh and ml within the range in which the
actual data fall between the high value Lh(t) and the low
value L1 (t) without fall.
At a determination step 1204, the prediction model
15 learning unit 111 determines with respect to every area in
which learning is possible whether or not learning about
area-specific prediction models has been made. If there
is any area in which no such learning has been made as yet,
the processing returns to the water distribution district
20 extraction step 1202. When learning has been completed
for every area in which such learning is possible, the
processing moves ahead to a generic prediction model
extraction step 1205.
At the generic prediction model extraction step 1205,
generic prediction model for the learning purpose.
At an area-specific prediction model extraction step
1206, the prediction model learning unit 111 extracts
every area in which sets of explanation variables to be
5 used by the generic prediction models to be learned can be
calculated, and extracts the area-specific prediction
model for the pertinent area out of these models.
On account of the nature of city water service which
I
usually uses pipelines installed underground for decades,
lo the pipeline information memory unit 121 and the survey
and repair history memory unit 122 do not keep on record
all information on every pipe. For this reason, in some
areas necessary information for calculating specific
explanation variables is not on record, and therefore
1s calculation may be impossible for some specific
explanation variables. In view of this problem, at the
area-specific prediction model extraction step 1206, the
prediction model learning unit 111 extracts only the areas
in which every set of explanation variables to be
20 calculated and their area-specific prediction models.
At a generic prediction model learning step 1207,
the prediction model learning unit 111 conducts learning
of the extracted generic prediction models. Learning
about the coefficients that determine generic prediction
extracted area-specific prediction models. The following
description assumes the use of the formula (2) as the
prediction formula for generic prediction models and the
use of the formula (1) as the prediction formula for each
5 area-specific prediction model.
The index for the extracted area and the areaspecific
prediction models are supposed to be p, the
coefficients of the prediction formula (1) for each areaspecific
prediction model, to be Lop, kp, DhOp, DlOp, mhp
l o and mlp, and the predicted values, to be Lp (t) , Lhp (t) and
Llp(t) . The value of the index s in the area p of
explanation variables shall be expressed as xsp, and that
of explanation variables in the area p, collectively as xp.
On this occasion, the prediction model learning unit
1s 111 so calculates coefficients aO, PO, as and ps as to
enable formulas (2d) and (2e) to well represent the
coefficients Lop and kp of the area-specific prediction
models. For this calculation, known art can be utilized,
such as the least squares method for instance. Further,
20 the prediction model learning unit 111, using the
prediction formula (2a) as described above, so determines
the smallest coefficients DhO, D10, mh and ml as to enable
the following two formulas to hold in every area p.
Lh(t, xp) 2 Lhp(t)
At a determination step 1208, the prediction model
learning unit 111 determines whether or not learning has
made for every generic prediction model. If there is any
generic prediction model for which no learning has been
5 made as yet, the processing returns to the step generic
prediction model extraction 1205. If learning has been
made for every generic prediction model, the processing
moves ahead to a transmission step 1209 for outputted
information. At the transmission step 1209 for outputted
lo information, the prediction model learning unit 111
transmits information on the learned prediction model to
the prediction model memory unit 131. At an end step 1210,
the prediction model learning unit 111 ends the processing.
Fig. 13 is a flow chart of processing by the leak
15 quantity predicting unit 112. Fig. 13 charts an operation
flow in which the leak quantity predicting unit 112
predicts the future leak quantity on the basis of
prediction models, chooses the predicted leak quantity
information according to the prediction model with the
20 smallest difference between the high and low values, and
transmits the choice to the survey schedule drafting unit
113 and the frame display unit 174. At a start step 1300,
the leak quantity predicting unit 112 starts processing.
At a reception step 1301 for inputted information,
k .=tk34lT3.-=~n-Z kL5 y:y-zpfijkj+i - " - - -
prediction model memory unit 131 prediction model
information prepared by the flow of Fig. 12. Further the
leak quantity predicting unit 112 receives as required
from the pipeline information memory unit 121, the survey
5 and repair history memory unit 122 and so forth
information needed for calculation of explanation
variables in each area.
At a water distribution district extraction step
1302, the leak quantity predicting unit 112 extracts one
l o water distribution district (area), which is to be the
object of leak quantity prediction. At a prediction model
extraction step 1303, the leak quantity predicting unit
112 extracts from the received prediction model
information all the prediction models applicable to the
15 extracted area.
At a prediction step 1304 using generic prediction
models, the leak quantity predicting unit 112 predicts the
leak quantity in extracted areas by using each generic
prediction model out of the extracted prediction models.
20 Here, the leak quantity prediction by the leak quantity
predicting unit 112 using generic prediction models means
calculation for the areas from which the values of the
required explanation variables were extracted and fitting
the calculated values of the explanation variables into
calculate the expected value, high value and low value of
the predicted leak quantity shown in Fig. 9.
At a determination step 1305, the leak quantity
predicting unit 112 determines whether or not any area-
5 specific prediction model applicable to the extracted
prediction models is included. If any is, the processing
moves ahead to a prediction step 1306 using the areaspecific
prediction model, or if none is, to a selection
step 1307 for prediction results.
10 At the prediction step 1306 using the area-specific
prediction model, the leak quantity predicting unit 112
predicts the leak quantity in the areas extracted by using
area-specific prediction model. Prediction of the leak
quantity here means, as in the case of generic prediction
15 models, processing to calculate the expected value, high
value and low value of the predicted leak quantity shown
in Fig. 9.
At a selection step 1307 for prediction results, the
leak quantity predicting unit 112 compares the leak
20 quantity prediction results using each of the extracted
prediction model, chooses the prediction model with the
smallest difference between the high and low values, and
adopts the predicted leak quantity information according
to the chosen prediction model as the predicted leak
38
At a determination step 1308, the leak quantity
I predicting unit 112 determines whether or not prediction
I has been processed for leak quantity prediction with
respect to every area. If there is any area for which
5 prediction has not yet been processed, the processing
returns to the water distribution district extraction step
1302. When every area has been processed for prediction,
the processing moves ahead to a transmission step 1309 for I
I outputted information. I
10 At the transmission step 1309 for, the leak quantity
predicting unit 112 transmits the calculated leak quantity
prediction information to the survey schedule drafting
unit 113 and the frame display unit 174. At an end step
1310, the leak quantity predicting unit 112 ends the
15 processing.
If the leak quantity predicting unit 112 finds a
plurality of prediction models applicable to a given area,
a prediction leak quantity with less uncertainty can be
provided by outputting predicted leak quantity information
20 according to a prediction model with the smallest
difference between the high and low values at the
prediction step 1306 using a step area-specific prediction
model.
To add, if pipeline renewal is scheduled during the I
the leak quantity predicting unit 112 may take the renewal
into consideration in its processing. For instance, if
the pipeline renewal would affect the calculation results
of the explanation variables for the area, prediction
5 according to a generic prediction model using the
calculation results of explanation variables based on the
information after the pipeline renewal can be outputted
for the period after the renewal.
Fig. 14 is a flow chart of processing by the survey
l o schedule drafting unit 113. Fig. 14 charts an operation
flow until the survey schedule drafting unit 113 transmits
to the survey scheduling memory unit 132 leak survey
scheduling information figured out with due consideration
given to mathematical optimization problem At a start
15 step 1400, the survey schedule drafting unit 113 starts
processing.
At a reception step 1401 for inputted information,
the survey schedule drafting unit 113 receives from the
leak quantity predicting unit 112 leak quantity prediction
20 information on each area prepared by the flow of Fig. 13,
and also receives cost information from the cost
information memory unit 125.
At a configuration step 1402 for the optimization
problem, the survey schedule drafting unit 113 configures
survey schedule. Constraints to the mathematical
optimization problem configured by the survey schedule
drafting unit 113 include, first, the restriction that the
total number of areas of simultaneous leak surveying
5 should not be greater than the prescribed number of leak
survey teams in any part of the period to which the
schedule applies and, second, the period of implementing a
leak survey in any one area should be secured continuously
for the whole length of time taken to leak surveying of
lo the pertinent area. The former takes account of the
personnel cost, and the latter, the efficiency of work.
Further, the objective function that minimizes the
mathematical optimization problem configured by the survey
schedule drafting unit 113 can be one that takes into
15 account, for instance, the sum of both the estimated cost
of leak surveying and the estimated uncertainty of the
leak quantity prediction pertaining to the cost of leak.
A specific case of the mathematical optimization problem
configured by the survey schedule drafting unit 113 will
20 be described below. As subscripts, an index a for areas
and an index to each month of the scheduled period (three
years for instance) will be used.
In a specific case of determining the period of
implementing a leak survey in each area during the
decision variables are binary variables y {- a, t), which
take a value 1 used for only the month t of starting the
leak surveying of the area a and 0 for all others.
To define the binary variables ~-{a, t), which take
5 a value 1 used for only the month t of starting the leak
surveying of the area a and 0 for all others, the
following constraint is obtained as the relationship
between y-{a, t) and z - {a, t).
[Formula 31
10
where
1 - a is the number of months (positive integer) taken
to carry out leak survey of the area a. This constraint
formula ensures that, as a length of time for implementing
15. leak survey in a given area, a continuous period can be
secured without fail as long as needed for completion of
leak survey of the whole pertinent area.
The constraint on the total number of areas in which
simultaneous leak surveying is implemented can be
20 expressed as:
[Formula 41
T is the number (positive integer) of leak survey
teams.
The objective function that minimizes the
mathematical optimization problem configured by the survey
5 schedule drafting unit 113 may be, for instance, the total
cost consisting of the sum of the leak cost and the survey
cost, for whose expression the predicted value CS of the
survey cost and the predicted value CW of the leak cost
are used.
10 F = CW + CS
Here, the survey schedule drafting unit 113, using
the leak survey cost C a for the area a on recorded - in the
cost information memory unit 125, is calculated to be, for
example :
15 [Formula 51
CS = CaCaCty,,t . .. ( 3 )
On the other hand, the survey schedule drafting unit
113 calculates the predicted leak cost CW on the basis of
both the expected value of the leak quantity and its
20 predicted uncertainty.
For instance, parameters expressing the uncertainty
between the high and low values calculated for predicted
leak quantity in each area are set, and estimation is made
with respect to the relationship between the parameters
parameters take on any random values in a prescribed
uncertainty set is calculated.
More specifically,
[Formula 61
5 Calculation is made of
where w is the marginal cost per unit quantity of leak
water on record in the cost information memory unit 125,
Lh - a (t: {y-{a, T)) is the high value of the predicted
l o leak quantity in the month t where a decision variable
y-{a, t) is prescribed, and
L1 - a (t: {y-{a, T))) is the low value of the predicted
leak quantity in the month t where the decision variable
y-{a, t) is prescribed.
15 Here, 6 - a is a parameter expressing uncertainty in
the area a, and the uncertainty set of parameters is
supposed to be a set in which a vector 6 comprising
arrayed parameters 6 - a in all the areas is used and the
norm of 6 is not greater than 1.
Further, Lh a and L1 - a are specifically calculated I
from the high and low values of the leak quantity received
from the leak quantity predicting unit 112 and determined
by the length of time elapsed after the leak survey and
period of leak survey scheduling on record in the survey
and repair history memory unit 122, and the implementation
timing of leak survey and repair during the period of leak
survey scheduling determined by the decision variables.
5 ~t a solution step 1403 for the optimization problem,
the survey schedule drafting unit 113 processes solution
of the optimization problem configured at the
configuration step 1402 for the optimization problem, and
covers the optimum solution obtained into leak survey
l o scheduling information. To the solution processing, a
known technique such as metaheuristics including a generic
algorithm or branch and bound method can be applied.
At a transmission step 1404 for outputted
information, the survey schedule drafting unit 113
15 transmits the calculated leak survey scheduling
information to the survey scheduling memory unit 132. At
an end step 1405, the survey schedule drafting unit 113
ends the processing.
A supplementary remark will be made on the predicted
20 value CW of the leak cost. Generally, leak survey
scheduling obtained by using the leak cost based only on
the expected value of the predicted leak quantity as the
objective function of a mathematical optimization problem
makes the cost (the objective function value of the
than the estimate at the time of scheduling if the real
leak quantity deviates from the predicted leak quantity.
The leak cost based only on the expected value is an
objective function using CWA defined by the following
5 equation.
[Formula 71
CWA = w C E La (t: { Y , T ] ~ ) . (5)
where
L - a (t: {y-{a, T))) is the predicted leak quantity in the
10 area a in the month t where a decision variable ~-{a, t)
is prescribed.
As prediction of a leak quantity always involves
uncertainty, the cost in a case of applying leak survey
scheduling the expected value is almost certain to be
1s greater than the objective function of the optimum
solution of the mathematical optimization problem. In
view of this disadvantage, a concept of robust
optimization is applied to the foregoing predicted value
CW, and the cost in a case of occurrence of a typical
20 error in the predicted leak quantity is calculated. By
drafting a leak survey schedule by using such an estimated
value, the cost occurring in real application of the leak
survey scheduling can be expected to prove smaller than
when an estimated value based only on the expected value
The predicted value CW of the leak cost is not
limited to what was described above. For instance, the
cost in an area where the uncertainty is significant can
be heavily weighted in calculating the leak cost. Also,
5 the configuration of the mathematical optimization problem
is not limited to the foregoing. For example, constraints
including a prescribed upper limit of the survey cost are
set, and the predicted value of leak cost can be used as
the objective function for minimization. As the objective
l o function for the mathematical optimization problem, the
leak cost of Formula 5 may be used, and the leak cost of
Formula 4 may only be calculated.
Fig. 15 shows frame display by the frame display
unit 174 regarding leak survey scheduling. A leak survey
1s scheduling display frame 1501 that the frame display unit
174 shows on a display or the like has a water
distribution block indication 1502, a cost indication 1503
and a tabulated leak survey scheduling indication 1504.
The frame display unit 174 shows in the water
20 distribution block indication 1502 the water pipeline
network, water distribution blocks, areas and so forth on
a map in in linkage with the GIs regarding areas which
are covered by.the schedule drafting by the leak survey
schedule drafting device 101. In this example, water
332 are shown.
The frame display unit 174 shows in the cost
indication 1503 the result of cost estimation by leak
survey scheduling. In the row of a survey cost 1541, the
5 cost of leak survey calculated by Formula (3) is shown for
instance. In the row of loss cost 1542, the sum of leak
costs calculated by Formula (4) is shown for instance. In
the row of total cost 1543, the total sum of the survey
cost and the lost cost is shown.
10 In the row of robust cost 1532, an evaluated cost
sum is shown, with the uncertainty of prediction being
taken into account. For example, for evaluation of the
loss cost, a value calculated by Formula (4) is shown. In
the row of average cost 1533 on the other hand, the result
15 of cost evaluation of leak survey scheduling in the case
of using the average predicted value (expected value) is
displayed. In estimating the loss cost for instance, a
value calculated by Formula (5); instead of that by
Formula (4), is shown.
2 0 The frame display unit 174 shows in the tabulated
leak survey scheduling indication 1504 a table of leak
survey scheduling as in the case of Fig. 11.
Fig. 16 shows frame display by the frame display
unit 174 regarding leak quantity prediction results. Fig.
display unit 174 regarding the result of leak quantity
prediction. A leak prediction display frame 1601
displayed by the frame display unit 174 has the water
distribution block display 1502, a tabulated leak
5 prediction display 1603 and a graphic leak water
prediction display 1604.
The frame display unit 174 shows in a tabular form
in the tabulated leak prediction display 1603 the trend of
prediction of leak quantity by the leak quantity
lo predicting unit 112. In the column of area information
1631, the area ID of the object of prediction is shown.
In the column of average predicted value 1632, the
expected value of predicted leak quantity is shown. In
that of high predicted value 1633, the high value of the
is predicted leak quantity is shown. In that of low
predicted value 1633, the low value of the predicted leak
quantity is shown. When the user of the leak survey
schedule drafting device 101 takes an action to alter the
point of time of the displayed object, the frame display
20 unit 174 alters the point of time of the displayed object
into a designated current (prediction timing) or future
time .
The frame display unit 174, in its graphic leak
water prediction display 1604, graphically shows the trend
3C; nf nf 1 r - r thr. 1 --lr -. -t: t.-
--- LFkJL -Uk- XPZ-Z---~~- r a A L L L y
predicting unit 112 as in Fig. 9. To add, a vertical line
1650 represents the point of drafting (present time) the
leak survey scheduling; the past is shown to the left of
the vertical line 1650, and the result of past leak
5 quantity prediction is shown to the right of the vertical
line 1650. When the user of the leak survey schedule
drafting device 101 takes an action to alter the area of
the displayed object, the frame display unit 174 so alters
the display as to show the result of leak quantity
l o prediction for the designated area.
To add, the area for which the leak survey schedule
drafting device 101 predicts the leak quantity or drafts a
survey schedule need not be all the areas of the water
pipeline network to which the pertinent action is applied.
1s For instance, survey schedule may cover only some areas
while prediction is made of all the areas.
As hitherto described, the leak survey schedule
drafting device 101 as configured in the foregoing manner
can draft leak survey schedules with high cost
20 effectiveness with limited available resources even in the
presence of uncertainty regarding leaks.
Although the above-described embodiment is intended
for survey schedule regarding leaks from the water service
network, the invention can also be usefully adapted to
drafting of a survey schedule for leaks from gas piping.
Further, the invention is not limited to the
embodiment described above, but includes various
modifications. For instance, the foregoing embodiment was
5 described in detail with an eye to helping better
understanding of the invention, which is not necessarily
limited to what has all the configuration features
described in the foregoing paragraphs.
Further, the configuration features, functions,
l o processing units and processing means of the foregoing may
as well be partly or wholly realized with hardware by, for
instance, designing them in integrated circuits. Also,
the foregoing configuration features, functions and other
aspects may also be realized with software by causing the
15 processor to interpret and execute programs to
implementing the respective functions. Information
including programs, tables and files to perform those
functions.can be stored in recording devices such as
memories, hard disks or solid state drives (SSDs) or
20 recording media such as IC cards, SD cards or DVDs.
Also, only those control lines and information lines I
considered necessary for effective description are shown,
but not all such lines present in the product are shown.
For practical purposes, it may be considered that i
one another
What is claimed is:
1. A drafting device for a leak survey scheduling,
drafting a leak survey scheduling to cover a plurality of
5 areas into which a water pipeline network is partitioned,
comprising:
a measurement information collecting unit for
collecting measurement information pertaining to water
flow rates from a flowmeter installed on a water pipeline
l o network and other measuring instruments;
a water consumption quantity memory unit for storing
information on water consumption quantities in the areas;
a leak quantity estimating unit for estimating water
leak quantities in the areas on the basis of the
15 measurement information and the water consumption quantity
information;
a pipeline information memory unit for accumulating
pipeline information including length information on the
water pipeline network in the areas;
2 0 a survey and repair information memory unit for
accumulating survey and repair information including the
implementation timing of leak surveying and pipeline
repairing in the areas;
a prediction model learning unit for generating
leak quantities in the areas on the basis of at least one
of the leak quantity information, the pipeline information
and the survey and repair information;
a leak quantity predicting unit for generating
5 predicted leak quantity information in the areas on the
basis of the prediction model information; and
a survey schedule drafting unit for drafting a leak
survey scheduling that prescribes the sequence of
implementing leak surveys in the plurality of areas on the
l o basis of the predicted leak quantity information,
wherein the leak quantity predicting unit generates
predictions of both the expected value of the leak
quantity and the uncertainty of the expected value of the
leak quantity as 'the predicted leak quantity information;
15 and
the survey schedule drafting unit drafts scheduling
by using a leak cost calculated on the basis of both the
expected value of the leak quantity and the uncertainty of
the expected value of the leak quantity.
2. The drafting device for leak survey scheduling
according to Claim 1,
wherein the leak quantity predicting unit generates
predictions of both the high value and the low value of
leak quantity as predictions of the uncertainty of the
the prediction model learning unit generates a
prediction formula to predict the high value and the low
value of the leak quantity and a coefficient of the
prediction formula as the prediction model information.
5 3. The drafting device for leak survey scheduling
according to Claim 2,
wherein the leak quantity predicting unit, if the
prediction model applicable to one area is available in a
plurality, chooses and generates the predicted leak
l o quantity information whose difference between the high
value of the leak quantity and the low value of the leak
quantity is the smallest out of applicable prediction
models.
4. The drafting device for leak survey scheduling
15 according to Claim 3,
wherein prediction models of the prediction model
information generated by the prediction model learning
unit are one of two types including an area-specific
prediction model applicable to one specific area and a
20 generic prediction model made applicable to a plurality of
areas by substituting explanation variables of areas
calculated on the basis of the piping network information
and the survey and repair information, and
the coefficients determining the generic prediction
earlier prescribed area-specific prediction model.
5. The drafting device for leak survey scheduling
according to Claim 4,
wherein the survey schedule drafting unit prescribes
5 the period of implementing a leak survey in each area
during the period for scheduling as leak survey scheduling
to be drafted, and
the survey schedule drafting unit drafts a leak
survey scheduling satisfying a first constraining
l o condition that the total number of areas in which leak
surveying is to be simultaneously implemented in each part
of the period for scheduling should not be greater than a
prescribed number of leak surveying teams and a second
constraining condition that the period of leak surveying
15 in any one area should be secured continuously for the
length of time required for completion of leak surveying
in the whole pertinent area.
6 . The drafting device for leak survey scheduling
according to Claim 5,
2 o wherein the survey schedule drafting unit drafts the
leak survey scheduling so as to minimize the total cost
including the sum of the leak cost and the survey cost,
7. The drafting device for leak survey scheduling
according to Claim 6 ,
9 C -
-- L P-U-U-k-L-lK 11
evaluating a leak cost which is one of the evaluation
indicators, sets a parameter expressing uncertainty
between the high value and the low value of the predicted
leak quantity in each area and performs evaluation in
5 terms of relationship between the parameter and the leak
cost.
8. The drafting device for leak survey scheduling
according to Claim 7,
wherein the prediction model learning unit uses, as
l o the explanation variables for use in learning the generic
prediction model, at least one item out of the number of
leak repairs in the area, the number of notified leak
repairs in the area, the number of water service pipes in
the area, the number of in-service years of the oldest
15 pipe in the area, the number of pipes in long use, and the
total length of the pipes in long use.
9. A drafting system for leak survey scheduling
comprising:
the drafting device for leak survey scheduling
20 according to Claim 1;
measuring instruments, including a flowmeter, for
transmitting measurement information to the drafting
device for leak survey scheduling; and
a survey terminal for transmitting survey and repair
by leak surveying, to the drafting device for leak survey
scheduling.
10. A drafting method for leak survey scheduling,
drafting leak survey scheduling in a plurality of areas
5 into which a water pipeline network is partitioned,
comprising :
a collecting step of collecting measurement
information pertaining to water flow rates;
a memorizing step of storing information on water
l o consumption quantities in the areas;
an estimating step of estimating water leak
quantities in the areas on the basis of the measurement
information and the water consumption quantity
information;
15 a pipeline information accumulating step of
accumulating pipeline information including length
information on the water pipeline network in the areas;
a survey and repair information accumulating step of
accumulating survey and repair information including the
20 implementation timing of leak surveying and pipeline
repairing in the areas;
a prediction model learning step for generating
prediction model information to predict the trend of water
leak quantities in the areas on the basis of at least one
and the survey and repair information;
a predicted leak quantity information generating
step of generating predicted leak quantity information in
the areas on the basis of the prediction model
5 information; and
a drafting step of drafting a leak survey scheduling
I that prescribes the sequence of implementing leak surveys
~ in the plurality of areas on the basis of the predicted
leak quantity information,
10 wherein the predicted leak quantity information
generating step generates predictions of both the expected
value of the leak quantity and the uncertainty of the
expected value of the leak quantity as the predicted leak
quantity information; and
15 the drafting step drafts a scheduling by using a
leak cost calculated on the basis of both the expected
value of the leak quantity and the uncertainty of the
expected value of the leak quantity.